A statistical approach to the spam problem

By Gary Robinson. 2003

147 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Combating Attacks and Abuse in Large Online Communities (2016, background): “email spam [179, 182], network intrusion attacks [129, 250] , and Internet worms [164].”
A survey and experimental evaluation of image spam filtering techniques (2011, methodology): “…techniques based on machine learning have been widely investigated by several authors, to implement textual content analysis (Sahami et al., 1998; Drucker et al., 1999; Androutsopoulos et al., 2000; Graham, 2002a; Robinson, 2003; Meyer and Whateley, 2004; Koprinska et al., 2007; Lai et al., 2009).”
Misleading Learners: Co-opting Your Spam Filter (2009, methodology): “Robinson's method [22] smooths P(S,i) through a convex combination with a prior belief x (default value of x = 0.”
SpamBayes: Effective open-source, Bayesian based, email classification system (2004, methodology): “Various techniques developed by Gary Robinson [3] alleviated these problems, including a Central Limit Theorem approach, which produced two internal scores, one for ham and one for spam.”
A review of machine learning approaches to Spam filtering (2009, methodology): “The individual meth- ods considered include naive Bayes, Fisher’s probability combination method (Robinson, 2003), v2 classification, decision trees and neural networks.”
Adversarial machine learning (2019, unknown): “classifies email using token counts [55].”
Man vs. Machine: Practical Adversarial Detection of Malicious Crowdsourcing Workers (2014, background): “email spam [35, 36], network intrusion attacks [22, 54], and Internet worms [29].”
Machine Learning in the Presence of an Adversary: Attacking and Defending the SpamBayes Spam Filter (2008, methodology): “SpamBayes classifies using token scores based on a simple model of spam status proposed by Robinson [20], based on ideas by Graham [9], together with Fisher’s method for combining independent significance tests [6].”
Open problems in the security of learning (2008, methodology): “[19] develop attacks for causing false positives against the statistical spam .lter, formalized by Robinson [24], which is used in the SpamBayes (spambayes.sourceforge.net), BogoFilter (bogo.lter.sourceforge.net), and SpamAssassin (spamassassin.apache.org) systems.”
Bagging Classifiers for Fighting Poisoning Attacks in Adversarial Classification Tasks (2011, background): “For instance, in intrusion detection skilled adversaries may inject poisoning patterns to mislead the learning algorithm which infers the profile of legitimate activities [11] or intrusions [15,4]; while in spam filtering adversaries may modify spam emails by adding a number of “good words”, i.”